resume-job-matcher
TypeScript CLI for AI resume screening: hard constraints gate candidates out before scoring, every score cites the resume span behind it, and an embedding pre-f
What is resume-job-matcher?
TypeScript CLI for AI resume screening: hard constraints gate candidates out before scoring, every score cites the resume span behind it, and an embedding pre-filter keeps large pools cheap. Claude, GPT, or OpenRouter.
Overview of resume-job-matcher
What it is
Resume Job Matcher is a TypeScript CLI that screens a folder of resume PDFs against a job description and identifies which candidates to interview. It uses a three-stage funnel: pre-filter (optional), gate (hard constraints), and score (anchored rating with evidence citations). It supports Claude, GPT, or OpenRouter as the LLM backend.
Key features
- Hard constraints from the job description (e.g., work authorization, on-site requirement, licenses) become pass/fail questions; failing a blocking constraint rejects the candidate before scoring.
- Every score cites the resume span that supports it, using an anchored 0–4 scale across seven criteria (language proficiency, education, years of experience, technical skills, certifications, soft skills, location).
- Optional embedding pre-filter (
--prefilter) keeps only thenresumes closest to the job by cosine similarity, reducing LLM calls for large pools. - Embedding cache is keyed on resume content, so a candidate is embedded once across all jobs.
- Output includes ranked results with gated-out and pre-filtered candidates listed separately, plus generated email (invite or rejection).
- Options include
--concurrency,--threshold,--no-email,--analyze-jd,--unify, and--no-analysis. - Every run ends with a token and cost line covering generation and embedding calls.
Use cases
Screening a folder of resume PDFs against a single job description, especially when hard constraints (e.g., work authorization, location) must be enforced before scoring. The pre-filter is useful for large candidate pools where running every resume through an LLM would be expensive. The tool is designed for recruiters or hiring managers who want traceable, evidence-based scores.
Getting started
Requires Node 20 or newer (tested on 22). No Python needed. bash npm install cp .env-example .env # add ONE key, for the mode you plan to use mkdir -p resumes # drop your candidate PDFs in here npm run match -- 'EXAMPLE job_description.txt' resumes --api openrouter Both positionals are optional, but their defaults (job_description.txt and src/) do not exist in a fresh clone, so pass them explicitly the first time. Modes (--api): openrouter (default), anthropic, openai.
License and pricing
The license is not stated in the repository metadata or README. The project is listed as open source (pricing type: open_source).
Based on the project's public README and repository metadata.